JOURNAL OF EXPERIMENTAL ZOOLOGY (MOL DEV EVOL) 302B:458–468 (2004)

Developmental Fate of the Mandibular in the , Lethenteron japonicum: Comparative Morphology and Development of the Gnathostome Jaw With Special Reference to the Nature of the Trabecula Cranii SHIGERU KURATANIn1, YASUNORI MURAKAMIw1, YOSHIAKI NOBUSADA2, RIE KUSAKABE1, and SHIGEKI HIRANO3 1Laboratory for Evolutionary Morphology, Center for Developmental Biology, RIKEN, Kobe, Hyogo 650-0047, Japan 2Department of Biology, Okayama University, Okayama 700-8530, Japan 3Department of Medical Technology, School of Health Sciences, Faculty of Medicine, Niigata University, 2-746, Asahimachi, Niigata 951-8518, Japan

ABSTRACT The jaw is a mandibular-arch derivative, and is regarded as the synapomorphy that defines the gnathostomes. Previous studies (Kuratani et al., Phil. Trans. Roy. Soc. 356:15, 2001; Shigetani et al., Science 296:1319, 2002) have suggested that the oral apparatus of the lamprey is derived from both the mandibular and premandibular regions, and that the jaw has arisen as a secondary narrowing of the oral patterning mechanism into the mandibular-arch domain. The heterotopy theory of jaw evolution states that the lamprey upper lip is a premandibular element, leaving further questions unanswered as to the homology of the trabecula in the lamprey and gnathostomes, and to the morphological nature of the muscles in the upper lip. Using focal injection of vital dyes into the cheek process core of lamprey , we found that the upper lip muscle and trabecula are both derived from mandibular mesoderm. Secondary movement of the muscle primordium is also evident when the expression of the early muscle marker gene, LjMA2, is visualized. A nerve-fiber labeling study revealed that the upper lip muscle-innervating neurons are located in the rostral part of the brain stem, where the trigeminal motor nuclei are not found in gnathostomes. We conclude that the lamprey upper lip is composed of premandibular ectomesench- yme and a lamprey-specific muscle component derived from the mandibular mesoderm innervated by lamprey-specific motoneurons. Furthermore, the lamprey trabecula is most likely equivalent to a mesodermally derived neurocranial element, similar to the parachordal element in gnathostomes, rather than to the neural-crest-derived prechordal element. J. Exp. Zool. (Mol. Dev. Evol.) 302B: 458–468, 2004. r2004 Wiley-Liss, Inc.

The origin of the vertebrate jaw, which is the Gregory, ’33; de Beer, ’31, ’37; Jarvik, ’80; Janvier, synapomorphy that defines the gnathostomes, has ’96; reviewed by Kuratani et al., 2001). From this been and remains an intriguing issue of vertebrate classical concept, several different lines of thought evolution. To address this question, the lamprey, a have developed. One theory suggests that the living jawless vertebrate, can serve as a model for velum, a pumping apparatus arising in the developmental and embryological comparisons embryonic oropharyngeal membrane (Kuratani with which to determine the event that gave rise to this evolutionary novelty (reviewed by Mallatt, 1996; and by Kuratani et al., 2001). wYasunori Murakami’s present address: Institut de Genetique et de The biting jaw of the gnathostomes is a Biologie Moleculaire et Cellulaire (IGBMC), CNRS/INSERM/ULP, CU de Strasbourg, BP 10142-67404 Illkirch Cedex, France derivative of the rostralmost , nCorrespondence to: Shigeru Kuratani, Laboratory for Evolution- the mandibular arch (Fig. 1A). Therefore, the ary Morphology, Center for Developmental Biology, RIKEN 2-2-3 Minatojima-minami, Chuo Kobe, Hyogo 650-0047, Japan. E-mail: evolution of the jaw is thought to have occurred [email protected] through the transformation of the rostralmost Received 31 March 2004; Accepted 3 May 2004 Published online 12 August 2004 in Wiley InterScience (www. branchial arch (Gegenbaur, 1898; Goodrich, ’30; interscience.wiley.com). DOI: 10.1002/jez.b.21011 r 2004 WILEY-LISS, INC. CRANIOFACIAL DEVELOPMENT IN THE LAMPREY 459 et al., 2001), was articulated dorsoventrally and ancestral developmental constraints, leading to differentiated into the jaw (reviewed by Carroll, the loss of morphological homologies, as is actually ’93; Janvier, ’96). However, there is no evidence to seen in the transition from agnathans to gnathos- clarify whether or not the gnathostome jaw arose tomes. Such a scenario fits the definition of from an ancestral differentiated velum, and the evolutionary novelty (Wagner and Mu¨ller, 2002). embryonic structure equivalent to the velum does The model of jaw evolution illustrated above not seem to persist into the adult gnathostome raises some questions of comparative morphology, (Kuratani et al., 2001). Other authors deny the which are still unresolved. Firstly, the muscles in velar origin of the jaw, and assert that the upper the upper lip are innervated by a branch of the and lower jaw elements are already found in maxillomandibular nerve (Johnston, ’05; Song and agnathan as a reduction of the Boord, ’93), which is inconsistent with the mor- premandibular mouth and simultaneous enlarge- phological concept of the ‘‘mandibular arch’’. ment of the mandibular arch to facilitate respira- Where these muscles originate is unknown. The tion, which led to the establishment of the premandibular mesoderm of the lamprey, which is gnathostome jaw (Mallatt, ’96). primarily located in the upper lip primordium In a series of previous studies, we proposed a (Fig. 1B), has been described as differentiating hypothesis referred to as the heterotopic theory of into the extrinsic eye muscle (Koltzoff, 1901), as in the vertebrate jaw (Fig. 1; see Kuratani et al., ’99, the gnathostomes. Although this idea is purely 2001; Shigetani et al., 2002). By examining the speculative, it is supported by the expression embryonic head of the lamprey, we found that the pattern of the Pitx homolog in the lamprey, as upper lip ectomesenchyme (see Horigome et al., reported recently (Boorman and Shimeld, 2002). ’99; Kuratani et al., ’99, 2001), which surrounds Secondly, if the origin of the upper lip ectome- the definitive cephalic mesodermal element (pre- senchyme is close to the gnathostome trabecular mandibular mesoderm), is located rostral to the cartilage, what is the morphological identity of the mandibular arch, unlike the upper jaw of gnathos- cartilage of the same name in the lamprey? tome embryos (Fig. 1B; Kuratani et al., 2001). The Alternatively, what is the developmental origin ectomesenchyme that participates in upper lip of the lamprey trabecula? The morphological formation is thus of premandibular origin and is identity and evolutionary origin of the trabecula not homologous to any mandibular derivatives, have long been debated, together with the ques- but is quite closely related to the tissue that tion of the ‘‘premandibular arch’’ in vertebrates differentiates into the prechordal region of the (Johnels, ’48; reviewed by Gregory, ’33; de Beer, neurocranium (Fig. 1A; ‘‘prechordal cranium’’; ’31, ’37; Janvier, ’96; and by Kuratani et al., ’97a, Couly et al., ’93) of the gnathostomes. Here, 2001; Fig. 1A). Experimental studies are necessary ‘‘premandibular’’ denotes the region rostral to to determine the morphological patterning of the the mandibular arch, not necessarily assuming the rostral head in the lamprey, to presence of a premandibular ‘‘arch’’; see Kuratani provide clues to the solutions of these questions. et al., 2001 and Shigetani et al., 2000, 2002 for The aim of the present study was to provide some premandibular ectomesenchyme. Therefore, topo- experimental data and further observations and graphical reorganization of the cephalic ectome- discussion on the developmental fate and behavior senchyme, or reassignment of the conserved of the mandibular mesoderm of the lamprey, molecular cascades involved in epigenetic interac- Lethenteron japonicum. By applying vital dyes tions to different subsets of the ectomesenchyme, and observing the gene expression that defines were assumed to have occurred in the transition subsets of developing muscles, we have collected from the agnathan to gnathostome stages of supplementary data to support the heterotopic evolution (Fig. 1B; Shigetani et al., 2002; see also theory of vertebrate jaw evolution. This study also Kuratani, 2003). This argument denies the mor- aimed to explain the morphological nature and the phological homologies of the upper and lower lips evolutionary origin of the lamprey trabecula. of the lamprey with the upper and lower jaws of the gnathostomes, respectively. The heterotopic theory reconciles the expression MATERIALS AND METHODS of homologous regulatory genes in nonhomologous Embryos tissues by heterotopyFchanges in developmental sites during evolution (Haeckel, 1875; reviewed by Mature male and female lampreys, Lethenteron Hall, 1998). Heterotopy implies the disruption of japonicum, were collected in a tributary of the 460 S. KURATANI ET AL.

Miomote River, Niigata, or purchased from a local Fig. 1. Origin of the gnathostome jaw. A. Top: generally fishery of Hokkaido, Japan, during the breeding assumed prototype of the vertebrate head with undifferen- seasons (early June) of 2001 through to 2003. The tiated array of pharyngeal arch skeleton. Middle: primitive eggs were artificially fertilized and kept in 10% gnathostome similar to extant shark possessing the mandib- Steinberg solution (Steinberg, ’57) at 161C. Em- ular arch (MA; pink) differentiated as a biting jaw, with the hyoid arch (HA; blue) as the jaw suspension, and the real bryonic stages were assessed morphologically branchial arches (BA; green) that are basically of the same according to the sequence established by Tahara morphological pattern. Note that both the upper and lower (’88) for L. reissneri, a brook lamprey species jaws are derived from a single arch (MA).There is a pair of rod- closely related to L. japonicum. like cartilages called the trabeculae (tr) in front of the mandibular arch as the anlage of the neural-crest-derived prechordal neurocranium. This cartilage used to be regarded Focal injections of DiI and DiO as a premandibular arch. Bottom: schematic representation of the pre-gnathostome ancestor in terms of the heterotopy Based on the method described by Shigetani theory (Shigetani et al., 2002), showing a morphology similar et al. (2000), focal injections of 1,1’-dioctadecyl- to that of the ammocoete larva of the lamprey. Three types of 3,3,3’,3’-tetramethylindocarbocyanine perchlorate morphological identities can be discerned, as in the gnathos- (DiI) and 3,3’-dioctadecyloxacarbocyanine perchlo- tomes (see Kuratani et al., 2001). In this animal, the oral rate (DiO) were made into the mandibular apparatus is formed of both the premandibular and mandib- ular ectomesenchyme; the upper lip (ulp) of this animal is mesoderm. Stage 21 or younger lamprey embryos derived from the ectomesenchyme equivalent that gives rise to (Tahara, ’88), in which the premandibular meso- the trabecula in the gnathostomes. B. Heterotopy theory of the derm has not yet invaded the cheek process and origin of the vertebrate jaw. In the lamprey , the the process contains only the mandibular meso- premandibular ectomesenchyme surrounding the premandib- derm, were used. Using a fine glass pipette, a ular mesoderm (pm) differentiates into the upper lip dorsal to the mouth (mo), and the mandibular ectomesenchyme around solution of dyes was injected into the left cheek the mandibular mesoderm (mm) into the lower lip (llp). These process using a Pneumatic PicoPump (PV830, portions of the ectomesenchyme commonly express Dlx1 World Precision Instruments, Inc., Sarasota, FL cognate (colored gray) to pattern the oral apparatus through USA). Embryos were then incubated for a further epigenetic interaction with ectodermally derived growth 2-7 days and were fixed with 4% paraformaldehyde factors (blue line). In gnathostome development, a similar interaction occurs only in the domain of the mandibular arch, (PFA) in phosphate-buffered saline (PBS) for 2 h and the premandibular ectomesenchyme does not express the at room temperature (RT). They were rinsed Dlx1 cognate. Dotted line indicates the morphological bound- briefly with 0.9% NaCl and immersed in 30% ary between premandibular and mandibular ectomesenchyme glycerol in distilled water. The embryos were (see Shigetani et al., 2000, 2002 for details). uj, upper jaw; lj, mounted on a depression slide glass and observed lower jaw. with a fluorescence microscope. Fig. 2. Dye injections into the lamprey mandibular Retrograde labeling of neurons mesoderm I. A. Scanning electron micrograph of a stage 21 lamprey embryo, left lateral view. The cheek process (cp) Rhodamine- or fluorescein-conjugated dextrans protrudes on the lateral aspect of the head. B. The surface (Sigma, St Louis, MO, USA) were injected into the ectoderm of the same stage embryo was removed to show the spinal cord or pharyngeal arches of the embryo to inside of the cheek process. Note that the rostral part of the label the reticulospinal or branchial motor neurons, process is filled with mandibular mesoderm (mm) and the caudal part with the endodermal first pharyngeal pouch (p1). respectively, according to the method described by No crest cells are visible on the surface of the mandibular Glover (’95). The injected embryos were incubated mesoderm. C. Transverse section of an embryo in which DiO at RT for 30 min to allow the dextran to label has been injected into the core of the cheek process and DiI neurons retrogradely. Embryos were then washed into the . Note that DiI-labeled cells are mostly with 10% Steinberg solution, and fixed in 4% PFA scattered in the superficial part of the head or in the surface ectoderm (arrowheads), whereas DiO-labeled cells are in the in PBS. The fixed specimens were dehydrated, and core (mesoderm) part of the cheek process (arrow). nt, neural clarified with a 1:2 mixture of benzyl alcohol and tube. D and E. Two days after DiI injection into the benzyl benzoate. Labeled neurons were then mandibular mesoderm. Two examples are shown as whole examined using a fluorescence microscope. mounts. In D, labeled cells are distributed in the upper lip (ulp), and in E, labeled cells are also seen dorsal and posterior to the upper lip. F-I. Four days after DiO injection into the Whole-mount in situ hybridization mandibular mesoderm. Four examples are shown as whole mounts. Labeled cells are distributed in various domains of Whole-mount in situ hybridizations were perfor- the upper lip and mandibular-arch derivatives. Note the med as previously described (Murakami et al., 2001). similarities in labeling patterns between D and H, and Fixed embryos were dehydrated and stored in between E and I. CRANIOFACIAL DEVELOPMENT IN THE LAMPREY 461

Fig. 3. Dye injections into the lamprey mandibular mesoderm II. Embryos are shown as whole mounts 7 days after DiI injections. Left lateral views except for E, which was sectioned transversely. A–D. Four embryos are shown as whole mounts. In three embryos (A, B, and D), labeling is heavy in the trabecular primordium (tr; red arrows) beneath and rostral to the eye (e). A fibrous pattern of labeling is seen in the upper lip (ulp) of all the embryos shown here. Labeling is also visible in the velum (vel) and the lower lip (llp), in a pattern similar to the muscle morphology. E. Transverse section of the embryo shown in A. Labeling is heavy in the mesenchyme lateral to the (n), corresponding to the site of trabecula development. oc, oral cavity; p1, pharyngeal pouch 1; nt, neural tube.

Fig. 4. Expression patterns of LjMA genes. A-C. Expres- sion of LjMA2 was detected by whole-mount in situ hybridiza- tion in a developing series of the lamprey. A. At stage 24, initial expression of LjMA2 is in the oral region. High levels of expression are seen in the upper lip muscle (ulm), which is located in the superficial portion of the cheek process, spanning the upper lip (ulp) and the dorsal part of the mandibular arch. B. In the mandibular arch, the velar muscle (vm) is visualized by LjMA2 gene expression. C. The upper lip muscle is now restricted within the upper lip. The first myotome (m1) has secondarily migrated rostrally, covering the upper lip and lower lip muscles laterally. LjMA2 expression at this stage delineates the basic anatomy of the lamprey muscles. D. LjMA1 expression at stage 28; the expression is restricted to the upper lip muscle. llp, lower lip; mo, mouth; m2, second myotome. 462 S. KURATANI ET AL. methanol at 201C. Specimens were treated over- (Kuratani et al., ’99). Furthermore, at this stage of night with a 1:5 mixture of hydrogen peroxide and development the migrating crest cells have not methanol, and were rehydrated in PBS containing arrived at this level (Horigome et al., ’99). There- 0.1% Tween 20 (PBT). After treatment with 0.2 N fore, it is possible to label only the mandibular HCl in PBT for 10 min at RT, the samples were mesoderm at this stage by injecting either DiI or digested with 10 mg/ml proteinase K (Sigma). They were postfixed for 20 min with 4% PFA in PBT containing 0.2% glutaraldehyde, then washed Fig. 5. Innervation and morphological patterns of the with PBT, and prehybridized in hybridization trigeminal nerves. A–C. Results of upper lip (ulp) and lower lip buffer (50% formamide, 5 saline sodium citrate (llp) labeling in lamprey larvae. With upper-lip labeling, the [SSC], 1% sodium dodecyl sulfate [SDS], 0.05 mg/ ventral part of the maxillomandibular ganglion (Vgl) and the ml total yeast RNA, 50 mg/ml heparin sulfate, posterior part of the trigeminal motoneurons (Vm) are labeled, and with lower-lip labeling, the dorsal part of the 5 mM ethylene diaminetetraacetic acid [EDTA]- ganglion and anterior motoneurons are labeled. Sensory and Na2, 0.1% CHAPS) for 1 h at 651C. The specimens motor somatotopies are shown in B and C, respectively. Dots were then incubated in hybridization buffer with in B indicate the contour of the maxillomandibular ganglion. 0.1 mg/ml DIG-labeled RNA probe for 48 h at e, eye; MHB, mid-hindbrain boundary. D. Comparison of 651C. After hybridization, the specimens were trigeminal-nerve patterning in lampreys and gnathostomes. Both in the lamprey and gnathostomes, the trigeminal nerve washed twice in 50% formamide, 5 SSC, and consists of the rostral part called the ophthalmic nerve (V1) 1% SDS for 30 min at 651C, and the solution was and the caudal maxillomandibular nerves (V2+3). In the substituted gradually with 10 mM Tris-HCl (pH lamprey, the dorsal component of the maxillomandibular 7.5) containing 0.5 M NaCl and 0.1% Tween 20 nerve is distributed in the upper lip located in the preman- (TST). RNaseA was added to a final concentration dibular part of the head. Sensory neurons innervating this area occupy the caudal part of the maxillomandibular gang- of 0.05 mg/ml and the specimens incubated for lion, which is not the case in the gnathostomes. The 30 min at RT. The samples were washed twice mandibular-mesoderm-derived upper lip (ulp) muscles are with 2 SSC in 50% formamide for 30 min at innervated by the caudal motoneurons in the hindbrain (pink 651C, twice in 2 SSC containing 0.3% CHAPS for in D). The caudal component of the maxillomandibular nerve 30 min at 651C, and twice in 0.2 SSC containing innervates the original mandibular-arch derivatives (ma), 1 which consist of the velum (vel) and lower lip (llp). In 0.3% CHAPS for 30 min at 65 C. For immunolo- gnathostomes, the oral apparatus differentiates from the gical detection, the embryos were blocked with mandibular arch, where the maxillomandibular part of the TST containing 0.5% blocking reagent (Roche) for trigeminal nerve is distributed. Note that the somatotopy of 60 min, and incubated with alkaline phosphatase the sensory neurons in the ganglion is the reverse of that in (AP)-conjugated anti-digoxygenin Fab fragment the lamprey, and that r4 does not contain any trigeminal 1 motoneurons. Also, note that the gnathostome maxilloman- (diluted 1:4000; Roche) at 4 C overnight. The dibular-oral system is composed of a subset of the elements specimens were washed five times for 60 min each seen in the lamprey. e, eye; MHB, mid-hindbrain boundary; in TST at RT. Alkaline phosphatase activity mn, lower jaw; mx, upper jaw; ph, pharynx; r1-4, rhombo- was detected with nitroblue tetrazolium (NBT)/ meres. 5-bromo-4-chloro-3-indolyl phosphate (BCIP) in NTMT (Roche). Fig. 6. Schematic representation of the developmental pattern of the lamprey oral apparatus. Top: early embryonic RESULTS pattern in which the cheek process contains only the mandibular mesoderm (mm) and the first pharyngeal pouch Labeling the lamprey mandibular (p1) surrounded by the trigeminal crest cells (TC). Middle: in mesoderm the cheek process at the following stage, the premandibular mesoderm (pmm) secondarily arises rostral to the mandibular In the early pharyngula of the lamprey around mesoderm, as the initial anlage of the lamprey upper lip (ulp). stage 21 (Tahara, ’88), a pair of obvious protru- Thus, the upper lip belongs to the premandibular domain (pm) of the lamprey head. Into this domain, the dorsal subset of the sions has developed on both sides of the rostral mandibular mesoderm grows to form the upper lip muscle. head. These structures are called ‘‘cheek pro- Here, the mandibular-arch domain (ma) is defined as the cesses’’ (Damas, ’44), and consist of expanding region occupied by the original mandibular mesoderm. Below: mandibular mesoderm and the first pharyngeal completed ammocoete oral apparatus is shown. The main part pouch endoderm (Fig. 2A, B; see also Kuratani of the mandibular mesoderm differentiates in situ into the muscles of the velum (vel) and the lower lip (llp), as well as et al., ’99, 2001). It is only at later stages that the the trabecula of the lamprey, which is equivalent to the premandibular mesoderm develops into this pro- gnathostome parachordal cartilages that develop lateral to the cess, as the lateral growth of the prechordal plate notochord (not shown). CRANIOFACIAL DEVELOPMENT IN THE LAMPREY 463

Fig. 7. Comparative morphology of the ‘‘trabecula’’. A and B. Morphology of the ammocoete cranial base by Johnels (’48). Ventral (A) and left lateral (B) views. The trabecula of the lamprey (trl) is a pair of cartilage rods beneath the brain, connected rostrally by a commissure cartilage at the base of the upper lip. This cartilage continues into the parachordal cartilage (pcl) posteriorly. At this stage, the lamprey trabecula is mostly prechordal in position. nc, nasal capsule; nt, notochord. C. Early development of the lamprey trabecula. The early primordium of the lamprey trabecula is located at the level of the notochord and mandibular-arch cartilage (ma). e, eye; ot, otic. A–D. Redrawn from Johnels (1948). D. Explanation of the ammocoete larva as an intermediate stage in the evolution of the gnathostomes, by de Beer (’37). Note that this author tried to find the trabecula in the perioral region, which does not represent the rostralmost part of the pharyngeal endoderm. ha, hyoid arch. E. Concept of the ‘‘prechordal cranium’’ proposed by Couly et al. (’93). Neural- crest-derived portion of the avian skull (stippled) is found rostral to the level of the notochord (nt). A large portion of the prechordal cranium differentiates from the paired ‘true’ trabecular cartilages that develop rostral to the notochord. Redrawn from Couly et al. (’93).

DiO fluorescent dye into the core of this process. For example, by injecting DiO into the cheek processes of embryos that had already been injected into the premigratory crest with another dye, DiI, we found that the presumptive mandib- ular mesoderm was selectively labeled with DiO, except for a small number of cells in the surface ectoderm. In contrast, a few DiI-labeled cells were found in a superficial position in the embryonic 464 S. KURATANI ET AL. head, apparently representing the cephalic crest myoblasts formed two distinct masses of cells: the cells (Fig. 2C). dorsolateral one that corresponds to the upper lip We injected the dyes into the stage 21 mandib- muscle, and the dorsoventrally extending medial ular mesoderm, and observed the distribution mass that later differentiates into the velar and of labeled cells 2, 4, and 7 days after injection lower lip muscles in the mandibular-arch domain (Figs. 2, 3). For example, two days after DiI (Fig. 4B). By stage 28, the basic anatomical injection, when the cheek process had begun to configuration of the lamprey trigeminal muscles develop a rostral process or the primordium of the was delineated by visualizing the expression of upper lip, the labeled cells were distributed in the this gene (Fig. 4C). The distribution and appear- cheek-process derivatives, including the upper lip ance of LjMA2-positive cells at this stage re- (Fig. 2D, E and data not shown). This pattern is sembled those of the DiO-labeled cells shown in consistent in embryos observed four days after Fig. 3. Therefore, considering these data together injection (Fig. 2F-I). In several cases, a superficial with the results of dye-labeling experiments, it is longitudinal strand of cells was labeled, which was likely that the presence of the upper lip muscles in also observed in an identical pattern in embryos the lamprey larvae is due to the secondary two and four days after injection (compare Fig. 2E migration of the subset of mandibular mesoderm and 2I). Seven days after injection (approximately cells that were primarily within the mandibular- stage 30; Fig. 3), labeled fibrous structures were arch domain, into the premandibular domain apparent in the upper and lower lips, as well as in established in the later embryo. the velum, representing the trigeminal-nerve- On the other hand, another member of the innervated muscle of the ammocoete larva (Fig. 3 LjMA gene family, LjMA1, is expressed in a more A–D; see below). High levels of labeling were often restricted subset of cranial muscles. According to a associated with cells located on both sides of the previous study, this gene is regarded as a marker notochord (Fig. 3E), corresponding to the trabe- for upper lip muscle (Fig. 4D), and in the cheek cular cartilage of this animal, according to the process of a stage 21 embryo, where no differentia- nomenclature of Johnels (’48). tion of the upper lip is visible, LjMA1 expression is upregulated only in the dorsal part of the mandibular mesoderm (Kusakabe et al., 2004). Expression patterns of genes encoding Therefore, the upper lip muscle might be specified muscle actin even before the morphological differentiation of Embryonic lamprey cranial muscles can be the cheek process into the upper lip and the confirmed by several different methods. By histo- mandibular-arch subdivisions in the early man- chemical analysis of acetylcholine esterase activ- dibular mesoderm, by the specific upregulation of ity, and immunostaining for tropomyosin, the LjMA1. The latter hypothesis requires confirma- fibrous structures labeled in the stage 30 embryos tion with further labeling studies, which will form and described above were found to be very similar part of our future research. to the trigeminal-nerve-innervated muscles of this animal (data not shown; see Kuratani et al., ’97b, ’99). In the present study, we found that members Sensory and motor innervation by the of the lamprey muscle actin gene family (LjMA1 larval lamprey trigeminal nerves and LjMA2) were specifically expressed in subsets of lamprey muscles and their precursors. The trigeminal-nerve-innervation pattern of the The expression patterns of LjMA2, an early developing lamprey was detected by the applica- marker of muscle development, were examined by tion of dextrans conjugated with different fluor- whole-mount in situ hybridization (Fig. 4A–C; a escent dyes or with biotin into scars made in the full description of the genes and their expression upper lip and mandibular-arch derivatives, includ- was published in Kusakabe et al., 2004). This gene ing the lower lip and the velum (Fig. 5). It was not is first expressed in the myotomes at stage 22, and possible to label the lower lip and the velum its initial expression in the cranial muscles was separately. To evaluate the sensory innervation, observed at stage 24, when a low level of transcript labeling was examined in the trigeminal ganglion was identified in the mesenchyme of the cheek (see Johnston, ’05 and Kuratani et al., ’97b), and process, in the shape of the early upper lip muscle, to evaluate the motor innervation, intramedullar and the rest of the trigeminal-nerve-innervated labeling was examined at the level of the rostral muscle group (Fig. 4A). At stage 25, presumptive hindbrain. CRANIOFACIAL DEVELOPMENT IN THE LAMPREY 465

When the dextrans were applied to the upper lip ’75; Kuratani, ’97). Our major findings are as domain, labeled cell bodies were seen in the follows. Firstly, labeling the lamprey mandibular ventroposterior subset of the trigeminal ganglion, mesoderm almost always results in the labeling of and in the posterior part of the trigeminal motor upper lip, lower lip, and velar muscles, as well as nucleus (Fig. 5A–C). The application of dextran to the mesenchymal condensation that represents the lower lip or velum resulted in the labeling of the trabecular primordium lateral to the noto- ganglia and motoneurons in a complementary chord. Secondly, the labeling pattern in the fashion: the dorsorostral part of the trigeminal presumptive muscle primordium is identical to ganglion and the rostral motoneurons were la- that observed with histological techniques and beled (Fig. 5A–C). Such labeling patterns are gene expression patterns (Figs. 3, 4). Thirdly, the consistent with the description by Song and Boord expression pattern of a member of the muscle- (’93) and Koyama et al. (’87). According to a actin-encoding gene family, LjMA2, is consistent previous observation, the upper-lip-innervating with the predicted migration of the mandibular- motoneurons appear to occupy a region corre- mesoderm-derived muscle primordium into the sponding to r4, which does not contain trigeminal premandibular domain (Fig. 4). Lastly, labeling motoneurons in gnathostomes (Fig. 5D; based on the developing trigeminal neurons suggests the the expression of the LjKrox-20 and LjPax6 genes; peculiar nature of the upper lip domain, compared see Murakami et al., 2004). with the patterns known in gnathostome develop- ment (Fig. 6). In the following discussion, evolu- DISCUSSION tionary and developmental establishment of the lamprey upper lip will be addressed mainly in the Due to technical difficulties, the developmental context of evolutionary developmental biology and dynamics of the lamprey head mesenchyme has comparative , to show that the ob- not been extensively studied. The precise morpho- served developmental patterns are consistent with logical distribution and premigratory mapping of our previous hypothesis of jaw evolution, and the cephalic crest cells have been described with finally to propose the evolutionary scenario of the scanning electron microscopy and whole-mount in biting jaw. In this context, the developmental situ hybridization of marker gene expression, as identity as well as the origin of the cartilage called well as with DiI labeling (Horigome et al., ’99; the trabecula will also be discussed. Kuratani et al., ’99; McCauley and Bronner- Fraser, 2002). However, no mapping data are Composite origin of the upper lip available for the organogenetic development of the mesoderm or ectomesenchyme in the lamprey. In all the vertebrate embryos so far studied, Classical descriptions in comparative embryology a continuous large ectomesenchyme called the are highly speculative, especially regarding the ‘‘trigeminal crest cells’’, can be found in the developmental fate of the mesoderm (Neal, 1897; rostral head, filling the mandibular arch and Koltzoff, ’01; Damas, ’44). The only available data the more rostral region, called the premandibular presented so far were derived from neural crest region (Fig. 1). The name of the trigeminal ablation experiments by Langille and Hall (’88), ectomesenchyme stems from the fact that the which showed the contribution of crest cells to the pattern of its distribution parallels that of the lamprey branchial cartilages (also see Newth, ’51, trigeminal nerve branches (reviewed by Kuratani ’56) and the neural crest specification along the et al., 2001). Therefore, the trigeminal crest cells neuraxis, using DiI (Horigome et al., ’99; Shigeta- can be anteroposteriorly subdivided into mandib- ni et al., 2002; McCauley and Bronner-Fraser, ular and premandibular domains, as the trigem- 2002). inal nerve is subdivided into the ophthalmic and In the present study, the developmental pat- maxillomandibular portions (compare Figs. 1 and terning of the mandibular mesoderm was first 5D). In gnathostomes, both the upper and lower observed experimentally, using cell labeling, ex- jaws develop from the mandibular domain, corre- amination of gene expression patterns, immuno- sponding to the distribution of axons belonging to staining, and the pattern of trigeminal axonogen- the second (maxillar) and third (mandibular) esis. It has often been exemplified in model branches of the trigeminal nerves (Fig. 5D; see animals, that the patterns of axonogenesis tend Kuratani and Horigome, 2000; Kuratani et al., to follow the migration and distribution of the 2001). Three major branches are also recognized mesenchymal components (Johnston, ’66; Noden, in the trigeminal nerve of the lamprey, including 466 S. KURATANI ET AL. the ophthalmic nerve homolog (Fig. 5D). Of these, 2004 and references therein). The craniofacial the posterior two branches, innervating the upper sensory innervation by the lamprey trigeminal lip and the velum + lower lip regions, have often nerve also appears peculiar because the somato- been called maxillar and mandibular nerves, topic organization of the trigeminal ganglion cells respectively (Johnston, ’05; Kuratani et al., ’97b; is reversed relative to that of the gnathostomes also see Goodrich, ’30 for general description). (Arvidson, ’77; Gregg and Dixon, ’73; Martin and Therefore, the patterning of the trigeminal nerve Dolivo, ’83; Noden, ’80; Scott and Atkinson, ’99). parallels that of the trigeminal ectomesenchyme, Therefore, the lamprey upper lip stands out as a and therefore the nomenclature and homology of truly exceptional morphological and embryological the nerve branches are tightly linked to the unit in the head of vertebrates. The gnathostome evolution of the jaw. oral apparatus appears to be constructed only Our previous studies have shown that the from the mandibular component of the lamprey patterning of the oral apparatus from the trigem- cheek process, consistent with the previous hy- inal ectomesenchyme proceeds through epithelial pothesis (FGF8 and BMP4 signaling cascades are interactions between the mesenchyme and the restricted to the mandibular domains of gnathos- surface ectoderm. In particular, the distribution of tomes, see Fig. 1B; Shigetani et al., 2002). growth factors plays a role in the prepatterning of For the reasons described above, it is unreason- the oral apparatus (Shigetani et al., 2000; Fig. 1B). able to call the upper-lip-innervating branch of the However, the distribution pattern of FGF8, lamprey trigeminal nerve the ‘‘maxillary nerve’’. which prefigures the central part of the mouth When compared with the developmental pattern opening, differs between the lamprey and gnatho- of the oral regions and the hindbrain in the stomes, resulting in a heterotopic shift in the gnathostomes, the lamprey appears to possess a epithelial-mesenchymal interactions (Fig. 1B). developmental program that does not exist in the Therefore, the homology between the ammocoete gnathostomes, which includes the upper lip pat- lips and gnathostome jaws is refuted; and at the terning mechanism, the rostrally migrating subset same time, the developmentally innovative nature of mandibular mesoderm, and the upper-lip- of the gnathostome jaw is confirmed (Kuratani innervating neurons (posterior part of the trigem- et al., 2001; Shigetani et al., 2002). In terms of the inal ganglion and motoneurons developing in r4). initial distribution of the ectomesenchyme, the If the gnathostome jaw has been acquired as an lamprey upper lip is identical to the premandib- apomorphic character, as is generally accepted, ular derivatives of the gnathostomes, such as the these evolutionary processes might have pro- nasal region and the trabecula (Figs. 1 and 5; see ceeded as the loss of the developmental module, Kuratani et al., 1997a). described above, in the ancestor of the gnathos- If the ammocoete upper lip has a premandibular tomes. On the contrary, it is possible that the identity, how can it possess trigeminal-nerve- developmental pattern in the gnathostomes repre- innervated muscles? Experiments labeling the sents the plesiomorphic state, irrespective of the mandibular mesoderm of the early lamprey fossil record (reviewed by Janvier, ’96). In this embryo, before the cheek process has differen- case, the upper-lip-related developmental mechan- tiated into the upper lip anlage or the premandib- ism should be regarded as newly invented in the ular domain, indicate that a part of the lamprey system, through the heterotopic shift of mandibular mesoderm secondarily grows ante- epigenetic interactions. riorly and laterally and migrates into the upper lip domain (Figs. 2–4, 6). No such muscles are Comments on the homology of the known in the gnathostomes, in which all the trabecula trigeminal-nerve-innervated muscles are restric- ted to derivatives of the upper and lower jaws. In the present study, labeling the mandibular It is not only the migration and gene specifica- mesoderm resulted in the labeling of the dense tions that are peculiar in the lamprey upper lip mesenchyme below the brain that corresponds muscles. These muscles also appear to be inner- to the site of lamprey trabecular development vated by a subset of trigeminal motoneurons (Fig. 3). In histological sections, the labeled cells located in r4 and rostral to the embryonic were also found lateral to the notochord, similar hindbrain (Fig. 5). This rhombomere is a segment to the site of the lamprey trabecula (Figs. 3E, of the hindbrain that never develops trigeminal 7B,C; Johnels, 1948). Such labeling patterns were motoneurons in gnathostomes (Murakami et al., observed consistently (eight of 12 successfully CRANIOFACIAL DEVELOPMENT IN THE LAMPREY 467 labeled embryos, 67%), as was the case with the ’48). This is perfectly consistent with the results of labeling of muscle tissues. Therefore, it is highly our labeling experiments (Fig. 7C). likely that this element actually represents the For the reasons described above, the lamprey trabecular cartilage of the lamprey, and that this homolog of the gnathostome trabecula, i.e., the cartilage is derived from a mesodermal compo- premandibular and prechordal part of the neuro- nent, the mandibular mesoderm. cranium, is found in the ectomesenchyme of the Historically, the morphological nature of the upper lip. As the result of a heterotopic shift in lamprey trabecula has been enigmatic (Fig. 7). ectoderm-ectomesenchyme interactions (involving This cartilage consists of a pair of rods beneath the FGF8 and BMP4), this part of the lamprey brain, which are connected rostrally by a trans- ectomesenchyme, which surrounds the preman- verse commissure at the base of the upper-lip- dibular mesoderm, has differentiated into a shape derived oral hood of the ammocoete larva, thus that resembles the gnathostome upper jaw providing the supporting tissue of the upper lip (Shigetani et al., 2002), again consistent with the (see Johnels, 1948, for anatomy; Fig. 7A, B). The present labeling data. resemblance between this structure and the gnathostome trabecula is obvious: both develop ACKNOWLEDGMENTS as a pair of rods beneath the brain and are united rostrally with their counterparts (see Kuratani We thank Dr. Kiyokazu Agata and Yasuyo et al., 2001). An important difference between Shigetani for their technical suggestions and them, however, is that the lamprey trabecula valuable discussion. This work was supported in develops at the level of the mandibular arch beside part by Grants-in-Aid from the Ministry of the notochord, whereas in the gnathostome, the Education, Science and Culture of Japan (Grant trabecula develops prechordally (Fig. 7C). For this numbers: 11309010, 11NP0201, and 11152225) reason, Johnels (’48) refuted the homology be- to S.K. tween these cartilages, if they are collectively called the ‘‘trabecula’’. De Beer (’31, ’37) also LITERATURE CITED refused to call this cartilaginous ring in the lamprey the ‘‘trabecula’’, but for another reason Arvidson B. 1977. Retrograde axonal transport of horseradish (Fig. 7D). For de Beer, the trabecula had to be a peroxidase from cornea to trigeminal ganglion. Acta premandibular structure, and he tried to find the Neuropathol. (Berl). 38:49–52. lamprey homolog of this cartilage in the upper lip Boorman CJ, Shimeld SM. 2002. Cloning and expression of a Pitx homeobox gene from the lamprey, a jawless vertebrate. (Fig. 7D). As far as the morphological identity of Dev. Genes Evol. 212:349–353. the ectomesenchymal domains is concerned, the Carroll RL. 1988. Vertebrate Paleontology and Evolution. latter idea is very close to our present conclusion W.H. Freeman and Co., New York. (also see below). Couly GF, Coltey PM, Le Douarin NM. 1993. The triple origin The prechordal position of the gnathostome of skull in higher vertebrates: A study in quail-chick chimeras. Development 117:409–429. trabecula implies its neural crest origin and also Damas H. 1944. Recherches sur le de´veloppment de Lampetra its chondrification in a notochord-free environ- fluviatilis L. contribution a` l0e´tude de la cephaloge´ne`se des ment. The concept of ‘‘prechordal cranium’’ verte´bre´s. Arch Biol Paris 55:1–289. proposed by Couly and colleagues (1993) perfectly de Beer GR. 1931. On the nature of the trabecula cranii. fits the embryonic development of the gnathos- Quart. J Microsc Sci 74:701–731. de Beer GR. 1937. The Development of the Vertebrate Skull. tome trabecula, which actually differentiates into Oxford University Press, London. the part of the cranium called ‘‘prechordal’’ Gegenbaur C. 1898. Vergleichende Anatomie der Wirbelti- (Fig. 7E). The data presented in the present study, hiere mit Beru¨cksichtung der Wirbellosen. Verlag von as well as the description by Johnels (’48), strongly Wilhelm Engelmann, Leipzig. suggest that what used to be called the lamprey Glover JC. 1995. Retrograde and anterograde axonal tracing with fluorescent dextran amines in the embryonic nervous ‘trabecula’ really has a mandibular mesodermal system. Neurosci Protocols 30:1–13. origin, and most likely corresponds to the para- Goodrich ES. 1930. Studies on the Structure and Development chordal cartilage of the gnathostomes, which of Vertebrates. McMillan, London. extends rostrally to the level of the forebrain Gregg JM, Dixon AD. 1973. Somatotopic organization of the (Fig. 7A–C). Such a rostral position can be trigeminal ganglion in the rat. Arch. Oral Biol 18:487–498. Gregory WK. 1933. Fish skulls: A study of the evolution of explained as the result of secondary elongation, natural mechanisms. Am Philos Soc 23:75–481. because the initial position of the trabecula is Haeckel E. 1875. Die Gastrea und die Eifurchung der Thiere. found at the level of the mandibular arch (Johnels, Jena Z Naturwiss 9:402–508. 468 S. KURATANI ET AL.

Hall BK. 1998. Evolutionary Developmental Biology, 2nd Ed. embryonic sea lamprey, Petromyzon marinus (L). Develop- Chapman & Hall, London. ment 102:301–310. Horigome N, Myojin M, Hirano S, Ueki T, Aizawa S, Kuratani Mallatt J. 1996. Ventilation and the origin of jawed verte- S. 1999. Development of cephalic neural crest cells in brates: A new mouth. Zool J Linn Soc 117:329–404. embryos of Lampetra japonica, with special reference to Martin X, Dolivo M. 1983. Neuronal and transneuronal the evolution of the jaw. Dev Biol 207:287–308. tracing in the trigeminal system of the rat using the herpes Janvier P. 1996. Early Vertebrates. Oxford Scientific Publica- virus suis. Brain Res 273:253–276. tions, New York. McCauley DW, Bronner-Fraser M. 2003. Neural crest Jarvik E. 1980. Basic Structure and Evolution of Vertebrates, contributions to the lamprey head. Development 130: Vol. 2. Academic Press, New York. 2317–2327. Johnels AG. 1948. On the development and morphology of the Murakami Y, Ogasawara M, Sugahara F, Hirano S, Satoh N, skeleton of the head of Petromyzon. Acta Zool 29:139–279. Kuratani S. 2001. Identification and expression of the Johnston JB. 1905. The cranial nerve components of Petro- lamprey Pax6 gene: evolutionary origin of the segmented myzon. Morphol Jahrb 34:149–203. brain of vertebrates. Development 128:3521–3531. Johnston MC. 1966. A radioautographic study of the migration Murakami Y, Pasqualetti M, Takio Y, Hirano S, Rijli F, and fate of neural crest in the chick embryo. Anat Rec Kuratani S. 2004. Segmental development of reticulospinal 156:143–156. and branchiomotor neurons in the lamprey: insights into Koltzoff NK. 1901. Entwicklungsgeschichte des Kopfes von evolution of the vertebrate hindbrain. Development Petromyzon planeri. Bull Soc Nat Moscou 15:259–289. 131:983–995. Koyama H, Kishida R, Goris RC, Kusunoki T. 1987. Neal HV. 1897. The development of the hypoglossus Organization of sensory and motor nuclei of the trigeminal musculature in Petromyzon and Squalus. Anat Anz 13: nerve in lampreys. J. Comp Neurol 264:437–448. 441–463. Kuratani S. 1997. Distribution of postotic crest cells in the Newth DR. 1951. Experiments on the neural crest of the chick embryo defines the trunk/head interface: Embryolo- lamprey embryo. J Exp Biol 28:247–260. gical interpretation of crest cell distribution and evolution of Newth DR. 1956. On the neural crest of the lamprey embryo. the vertebrate head. Anat Embryol 195:1–13. J Embryol Exp Morphol 4:358–375. Kuratani S. 2003. Evolution of the vertebrate jawFhomology Noden DM. 1975. An analysis of the migratory behavior of and developmental constraints. Paleontol Res 7:89–102. avian cephalic neural crest cells. Dev Biol 42:106–130. Kuratani S, Horigome N. 2000. Development of peripheral Noden DM. 1980. Somatotopic organization of the embryonic nerves in a cat shark, Scyliorhinus torazame, with special hick trigeminal ganglion. J Comp Neurol 190:429–444. reference to rhombomeres, cephalic mesoderm, and distri- Scott L, Atkinson ME. 1999. Compartmentalisation of the bution patterns of crest cells. Zool Sci 17:893–909. developing trigeminal ganglion into maxillary and mandib- Kuratani S, Matsuo I, Aizawa S. 1997a. Developmental ular divisions does not depend on target contact. J Anat patterning and evolution of the mammalian viscerocranium: 195:137–145. Genetic insights into comparative morphology. Dev Dyn Shigetani Y, Nobusada Y, Kuratani S. 2000. Ectodermally 209:139–155. derived FGF8 defines the maxillomandibular region in the Kuratani S, Ueki T, Aizawa S, Hirano S. 1997b. Peripheral early chick embryo: epithelial mesenchymal interactions in development of the cranial nerves in a cyclostome, Lampetra the specification of the craniofacial ectomesenchyme. Dev japonica: Morphological distribution of nerve branches and Biol 228:73–85. the vertebrate body plan. J Comp Neurol 384:483–500. Shigetani Y, Sugahara F, Kawakami Y, Murakami Y, Hirano Kuratani S, Horigome N, Hirano S. 1999. Developmental S, Kuratani S. 2002. Heterotopic shift of epithelial- morphology of the cephalic mesoderm and reevaluation of mesenchymal interactions for vertebrate jaw evolution. segmental theories of the vertebrate head: Evidence from Science 296:1319–1321. embryos of an agnathan vertebrate, Lampetra japonica. Dev Song J, Boord RL. 1993. Motor components of the trigeminal Biol 210:381–400. nerve and organization of the mandibular arch muscles in Kuratani S, Nobusada Y, Horigome N, Shigetani Y. 2001. vertebrates. Acta Anat 148:139–149. Embryology of the lamprey and evolution of the vertebrate Steinberg M. 1957. A nonnutrient culture medium for jaw: Insights from molecular and developmental perspec- amphibian embryonic tissues. Carnegie Institution of tives. Phil Trans Roy Soc 356:15–32. Washington Year Book 56:347–348. Kusakabe R, Takechi M, Tochinai S, Kuratani S. 2004. Tahara Y. 1988. Normal stages of development in the lamprey, Lamprey contractile protein genes mark different popula- Lampetra reissneri (Dybowski). Zool Sci 5:109–118. tions of skeletal muscles during development. Mol Dev Evol Wagner GP, Mu¨ller GB. 2002. Evolutionary innovations 302B:121–133. overcome ancestral constraints: A reexamination of char- Langille RM, Hall BK. 1988. Role of the neural crest in acter evolution in male sepsid flies (Diptera: Sepsidae). Evol development of the trabeculae and branchial arches in Dev 4:1–6.